Researcher honored for AI brain implant that helps restore speech in ALS
Experimental interface translates brain signals into spoken language in real time
Written by |
- An AI-powered brain implant helped an ALS patient regain speech by decoding brain signals.
- This neuroprosthesis translates brain activity into spoken language, even capturing intonation.
- The neoriscientist who led its development earns 2026 Chen Institute and Science Prize for AI Accelerated Research.
An experimental brain-computer interface that helped a man with amyotrophic lateral sclerosis (ALS) communicate again has earned a neuroscientist at the University of California, Davis, the 2026 Chen Institute and Science Prize for AI Accelerated Research.
Sergey Stavisky, PhD, an associate professor in the university’s Department of Neurological Surgery, was recognized for leading the development of an artificial intelligence (AI)-powered speech neuroprosthesis that translates brain signals into spoken language in real time, helping people to communicate after they lost the ability to speak.
The prize, awarded by the Tianqiao and Chrissy Chen Institute in partnership with the journal Science and the American Association for the Advancement of Science, honors researchers whose work is advancing science through AI.
“Stavisky developed an AI-based speech neuroprosthesis with immediate and transformative practical impact,” Yury V. Suleymanov, senior editor at Science, said in a press release from the Chen Institute. “It restored communication for a paralyzed patient with ALS with over 99% word accuracy, enabling the patient to express over 2.7 million words over two years using only brain signals. His team achieved real-time voice synthesis, allowing the patient to modulate intonation and even sing.”
Decoded text spoken aloud using digital version of the person’s voice
ALS gradually damages the nerve cells that control voluntary movement. As the disease advances, weakness in the muscles of the tongue and jaw can make speech difficult or impossible. While eye-tracking devices and other assistive communication systems can help, they often become increasingly difficult to use as muscle weakness worsens.
Recognizing that restoring communication is one of the most urgent unmet needs for people living with ALS, Stavisky and his team set out to develop a brain-computer interface that could restore speech by decoding the brain signals generated when a person simply attempts to talk.
The system relies on four tiny electrode arrays implanted in the brain over a region that’s critical for speech. Together, the arrays contain 256 electrodes that record patterns of brain activity when a person tries to speak. An AI-powered decoder translates those signals into text in real time, which is then spoken aloud using a digital version of the person’s own voice created from recordings made before ALS took away their ability to speak.
Goal is to develop system for widespread clinical use
In 2024, the team reported that the system allowed Casey Harrell, a 45-year-old man with ALS whose disease had made it nearly impossible to speak, to communicate again. About a month after surgery, the system decoded his words with 99.6% accuracy when he was asked to say sentences using a 50-word vocabulary.
After the vocabulary was expanded to about 125,000 words, representing most of the English language, further training allowed the system to maintain 97.5% accuracy. Harrell was able to hold self-paced conversations at about 32 words per minute.
The following year, the researchers advanced the technology even further with a brain-to-voice neuroprosthesis that bypasses text altogether.
Instead of first converting brain signals into written words, the AI directly generates speech from neural activity, producing a personalized voice with a delay of only about 30 milliseconds — close to the natural delay between speaking and hearing one’s own voice. The system also captures natural features of speech, including changes in intonation, allowing users to ask questions, emphasize words, and even sing.
Dr. Stavisky set out to solve a problem many in his field considered unsolvable, and in doing so gave people back something profoundly human: their own voice.
In their latest study, the researchers showed that Harrell had independently used the brain-computer interface at home for nearly two years. Over that time, Harrell used the system for more than 3,800 hours, communicating more than 183,000 sentences — nearly 2 million words — at an average speed of about 56 words per minute. He used it to talk with family and friends and continue working full time despite severe paralysis caused by ALS.
Stavisky’s long-term goal is to create a surrogate voice so natural that listeners would not realize it was AI-generated, while developing smaller, fully implanted systems that could one day move from research studies into everyday clinical care.
“Ten years ago, Tianqiao Chen and I founded the Chen Institute around a single question, not an answer: how does the brain give rise to intelligence? Where that question would lead was impossible to know, but that is the nature of real discovery,” said Chrissy Luo, cofounder of the Tianqiao and Chrissy Chen Institute. “Dr. Stavisky set out to solve a problem many in his field considered unsolvable, and in doing so gave people back something profoundly human: their own voice.”
Leave a comment
Fill in the required fields to post. Your email address will not be published.